According to Microsoft’s Global Online Safety Survey, released this month and conducted across 15 countries last year, the adoption of artificial intelligence by the general public has grown significantly despite the fears it raises. For example, 51% of people surveyed have used or experimented with generative AI in the past three months, an increase of 13 points over the previous year.
Across all age groups, people from India are the biggest fans of generative AI, with 80% having already tried the tech. They are followed by those in Singapore (66%), South Africa (60%), Brazil (53%) and the United States (50%).
Among those surveyed, the French are the least interested, with only 35% saying they have already tested these tools. However, 88% of all respondents say they are concerned about the risks posed by this technology, particularly in terms of data protection and image manipulation.
The study also reveals that 66% of respondents have been exposed to a risk related to generative AI in the past year. The most frequent dangers include misinformation or disinformation (47%), hate speech (33%), access to violent content (27%) and cyberbullying (21%).
Exposure to content relating to self-harm and suicide (13%) and deepfake pornography (12%) is also a significant risk.
Among the most exposed are teenagers: when faced with an online risk, the majority of them change the content, block a contact or mute the sound (74%). They then do not hesitate to talk to someone about their experience (66%).
Only 37% report the incident, believing it would change nothing and that it would have no consequences for the perpetrator.
Overall, the gap between the risks actually experienced by teenagers and their parents’ assessment of them is gradually narrowing: last year, 63% of parents believed their children had been exposed to an online risk, while 69% of teenagers said they had actually experienced it.
In terms of usage, young people are keen on the practical applications of AI, such as translation or homework help, but concerns are emerging about possible misuse. For example, 82% of parents fear their children will use AI to cheat at school, and are just as likely to fear that their kids will come across harmful content or deepfakes featuring acquaintances.
The study also highlights a lack of ability when it comes to identifying AI-generated content: only 38% of those surveyed were able to distinguish a real picture from an artificially generated image in a test.